Discussion

During FAST implementation at a general hospital in Lima, Peru, patients with pulmonary TB were more likely to start TB treatment during hospitalization and experience shorter time from admission to treatment than a cohort of historical patients hospitalized prior to FAST implementation. Our findings extend the accumulating body of evidence linking FAST to decreased time to TB treatment by demonstrating that the strategy can accelerate care outside of the specialized TB hospital. ^{Reference Le, Nguyen and Tran27,Reference Sani, Mustapha and Jumoke30}

Our study uniquely assessed how FAST could be used to enhance indicated treatment, that is, a regimen tailored to DST results and following local guidelines. This distinction is pertinent because prompt delivery of effective treatment based on DST is paramount to interrupting TB transmission. ^{Reference Riley, Mills and Nyka23,Reference Riley, Mills, O’Grady, Sultan, Wittstadt and Shivpuri24} Rapid DST also allows scarce isolation facilities to be prioritized for patients with DR-TB. These effects are particularly important in settings like Peru, where empiric first-line TB therapy cannot be assumed to be effective in newly diagnosed patients given the high prevalence of DR-TB. ^{1,Reference Barrera, Livchits and Nardell20}

Universal TB screening during routine care for all hospitalized patients is a key intervention in high TB–incidence settings. In 41 months at HNHU, under the FAST strategy >67,000 patients were screened for symptoms. Screening took place primarily in the emergency department, where the risk for TB transmission is high. ^{Reference Escombe, Huaroto and Ticona31} HNHU serves a much broader population of patients than the chest or TB hospitals where FAST was previously pilot tested. ^{Reference Nathavitharana, Daru and Barrera32} Nevertheless, the number of detected cases at HNHU corresponds to an extremely high TB prevalence: nearly 2,400 of 100,000 symptomatic individuals and >6,000 of 100,000 among those with symptoms who consented to participate. This burden of TB disease is comparable to that found in prisons and other institutional settings in which routine screening for active disease is recommended to reduce transmission and improve outcomes. ³³

The FAST screening protocol at HNHU evolved over the course of the study. The final iteration queried patients about any duration of cough and/or other risk factors for TB. Emerging data, however, demonstrate that TB in asymptomatic patients may also contribute to transmission. ^{Reference Herdman, Tovar and Wingfield34} Future research should examine the feasibility and performance characteristics of a FAST strategy that starts with a chest radiograph (irrespective of symptoms) to identify additional individuals with transmissible TB disease.

Xpert testing provided rapid and highly sensitive information about the presence of Mycobacterium tuberculosis. The increased sensitivity of Xpert testing led to the diagnosis of TB in 26 FAST patients (35%) in whom smear microscopy was negative. Xpert testing was also likely a prime factor in decreasing the overall time to TB treatment initiation, as has been demonstrated in other settings. ^{Reference Ershova, Volchenkov and Somova35}

The use of Xpert testing as part of the FAST strategy additionally led to the identification of 18 patients (24%) with RR/MDR-TB in the prospective cohort relative to only 4 (5.3%) with MDR-TB in the historical cohort. There was no reported secular increase in the prevalence of MDR-TB between these contiguous study periods. It is more likely that some patients with RR/MDR-TB were missed in the historical cohort because sensitive, rapid DST was not widely performed. In settings with a high prevalence of RR/MDR-TB, universal Xpert testing for all patients being evaluated for pulmonary TB should be the standard of care to reduce missed RR/MDR-TB diagnoses.

Relative to the historical controls, a higher percentage of patients with pulmonary TB in the FAST cohort received the indicated treatment within 2 days of diagnosis. Although GenoType was used for DST in nearly 60% of the historical cohort, it was not performed in real time; samples were batch-processed, thereby prolonging the interval to results. In contrast, under the FAST strategy, regimens were was informed by near real-time Xpert (and GenoType) results. Thus, it was possible to avoid empiric treatment and to promptly start regimens informed by results from rifampin (and isoniazid) resistance testing. The 2-day benchmark for indicated treatment should be achievable for most patients hospitalized with TB, even for those with DR-TB, given the advent of rapid molecular tests like Xpert to guide regimen composition (especially with development of cartridges that test resistance to isoniazid and fluoroquinolones, as well as rifampin). We suggest treatment initiation within 2 days as a minimum target for patients hospitalized with TB. Future efforts should try to reduce that time even further.

Shortening the time to treatment for patients with TB is critical to decreasing TB transmission by reducing the duration of infectiousness. ^{Reference Nathavitharana, Lederer, Tierney and Nardell36} In settings with a high burden of MDR-TB, increasing the speed of effective treatment for MDR-TB can lead to decreased transmission of MDR-TB, as previously demonstrated in a FAST pilot test in Russia. ^{Reference Miller, Livchits and Ahmad Khan37} In future research, we will assess the effect of FAST on TB transmission in hospital workers at HNHU.

Although the FAST strategy is conceptually straightforward, successful implementation depends on overcoming practical barriers to coordinated care in a complex hospital environment. The use of trained FAST assistants to facilitate the TB care cascade likely helped to overcome these barriers at HNHU. One potential mechanism may have been the expedited transfer of sputum samples to the laboratory and communication of diagnostic testing results to clinicians. Additionally, FAST assistants referred patients to HNHU physicians for consideration of therapy; if necessary, they also moved medications from the pharmacy to the ward to ensure that treatment was started as soon as possible. We expect that FAST can be adapted to local environments by training existing staff with modest additional inputs from healthcare systems. FAST implementation outside the study conditions has been demonstrated in TB hospitals in Bangladesh and Russia. ^{Reference Nathavitharana, Daru and Barrera32,Reference Miller, Livchits and Ahmad Khan37}

Our study has several limitations. First, comparing FAST patients to a historical cohort potentially introduces selection bias. Because the historical cohort comprised only patients with positive smear microscopy, we may have underestimated the impact of FAST. If patients with smear-negative TB were included in the historical control, time to TB diagnosis and treatment would likely have been longer because these relied on batched GenoType or mycobacterial culture. Second, the relatively short inclusion period for the historical cohort precluded detection of seasonal variations or other time trend effects. Third, 1,888 eligible patients (60.4%) opted not to participate. Their inclusion would have more than doubled the number of patients tested and/or treated and might have overwhelmed the system, diluting the effects of the FAST implementation. It is also possible, however, that the benefits of the FAST strategy might be more pronounced with a larger intervention sample size. Notably, if the FAST strategy became routine, it would not require consent and participation would be universal, potentially leading to benefits for more patients in the form of faster time to TB diagnosis and treatment.

In conclusion, FAST improves the likelihood and speed of DST-informed treatment for patients hospitalized with TB. This impact is important for TB transmission control because it reduces the time during which hospital transmission can occur. ^{Reference Nathavitharana, Lederer, Tierney and Nardell36} Faster TB treatment also represents a measurable improvement in quality of care for patients, meeting an urgent need in countries with a high burden of TB. ^{Reference Cazabon, Alsdurf and Satyanarayana38} Broader roll out of the FAST strategy in general hospitals in TB-endemic settings should be considered as part of a comprehensive TB transmission control approach that includes environmental controls and personal protective equipment, making these facilities safer for visitors, staff and patients.